By Giovanni DeMeo, Vice President, Global Marketing and Analytics
Predictive analytics gives retail marketers the power to see into the future and know with almost complete certainty exactly what products shoppers will purchase, where they will make those purchases and how their preferences will change in the short and long term. Of course, all of this can be done simply by clicking a button–or at least that’s what many of us are being lead to believe. While this is an extreme exaggeration, marketers can leverage the knowledge of analytical experts to use data from shopping transactions, loyalty cards, social media, mobile devices and a myriad of other sources to gain insight into how shoppers will likely make certain purchasing decisions, how they will respond to specific events (such as weather, holidays, etc.), and when and where those decisions will occur. If used effectively, progressive retailers and CPGs can use predictive analytics to better serve their customers, improve return on investment (ROI) from marketing initiatives and outperform their competitors.
In its simplest terms, predictive analytics uses a combination of data sets and mathematical formulas to predict the likelihood of some future event or action. In the case of retail, it can help predict how a shopper will likely act or react in the future. With the right data, technology and knowledge experts, predictive analytics can be (and in many instances already is) the single largest contributor to the growing gap between industry leaders and everyone else.
While virtually every retailer and CPG is already using data analytics in some form, the vast majority have the opportunity to raise the bar on how they’re using available data. For example, it is common practice to look at purchase behavior and target shoppers with marketing materials for products similar to those items purchased in the past (or commonly related items, like milk and cereal). But “in today’s hyper-competitive market, it is not enough for retailers to understand customer behavior—to be successful, retailers also must develop the ability to predict upcoming consumer actions,” says Marcy Patzer, senior director of retail strategy for Scala, Inc., in a recent Retail TouchPoints article. With predictive analytics, it is possible to send customized marketing communications based on behavioral patterns that indicate what future purchases will likely be, even if on the surface those items appear to be completely unrelated. For example, data mining may reveal that people who buy a certain brand of coffee creamer are also more likely to purchase canned soups. Discovering this connection would then allow marketers to cross-promote the two seemingly unrelated items.
A real game-changer in this realm is predictive modeling, which applies logic to the data by integrating historical patterns and external data to predict the future, improve decision making, optimize business performance and improve ROI. This is where the art of predictive analytics comes in. To craft models that can accurately forecast future behaviors and events, data scientists must first determine which combinations of attributes and variables are the most predictive of certain behaviors. Once these are identified, they can then be overlapped and applied to larger sets of data to anticipate everything from when a shopper is most likely to buy a particular product, to what brand he or she is most likely to favor, to whether a special offer will influence his or her decision.
One challenge for retailers is determining how to develop the resources necessary to take advantage of this proverbial goldmine. Just as retailers don’t raise their own cows in order to sell milk in their stores, high-level analytics is a competency they can’t expect to develop entirely own their own. Instead, they should look to develop a partnership with experts who specialize in consumer data and mathematical modeling. The first and most cost-effective option is to make use of the investment that many of their partners have already made. For example, Interactions includes these analytical services with its in-store demonstration, insights and other services in order to ensure our partners continue to be best-in-class retailers and to outperform their competition. The second option is to contract with a separate company that sells software and modeling services. The decision on which approach to take is most often dictated by a combination of factors, including cost, required human resources and overall corporate strategy.
Once the right resources are in place, “predictive analytics lets grocers peer into the future and find ways to better engage customers and meet their needs with a localized merchandise assortment, relevant promotions and personalized communications,” says Diana McHenry, global sales manager for SAS, in an interview with Grocery Headquarters. For example, it can help with promotional bundling and time-of-day optimization by determining what products should be sold together and when certain products sell best. This type of information can be leveraged to help customize in-store product sampling events for the exact days and times shoppers are looking for those items—going beyond the industry norm of simply scheduling events at the same time every day, regardless of the item or target consumer. It can also help retailers manage inventory by more accurately forecasting demand, thereby avoiding out of stocks and reducing shrink by minimizing product spoilage.
The bottom line is that predictive analytics enables retailers to truly “know their customer”—down to individual wants, needs and preferences. Gone are the days of being able to stay competitive using backward-looking, intuition-based decision making that has been the mainstay for decades. Future sales depend on knowing what your shoppers want—without even asking them. So to succeed in today’s marketplace, retailers and CPGs need to fully embrace—and trust—the new data-driven analytics that are the undeniable future of retail.